Feature Requests

Import skills in .md format to perform different actions and card interactions on the Chat section. Import as many skills as you need in a personal library and invoque with @ or as default action with a new card instead of creation a brief summary.

Currently, Recall automatically tags new content, but there is no way to retroactively update existing cards when new tags are created or when the tag structure changes. This leads to an inconsistent knowledge base where older cards miss out on newer, more relevant tags. I am proposing a "Tag Audit" feature (or a dedicated button) that would: Scan existing cards to apply newly created tags where relevant. Suggest global reorganizations to merge duplicates or fix "messy" hierarchies. Apply multiple relevant tags instead of just one, ensuring cards are correctly cross-referenced. This would ensure the knowledge base stays organized and up-to-date as the user's tagging system evolves over time.

Introduce a mid-tier subscription plan that excludes access to top-tier frontier models, offering instead a curated selection of lower-cost/lower-tier models (e.g., Sonnet, GPT-5.6 Turbo). Proposed Solution Add a new mid-tier plan positioned between the current entry-level and top-tier offerings. This plan would provide access to a subset of "lower level" models (e.g., Sonnet, GPT-5.6 Turbo) rather than the full model roster. Price the tier to reflect the reduced model access, making it more cost-effective for users whose workflows don't require frontier-model performance. Rationale Encourages better cost-to-value alignment for users with lighter or more routine use cases. Allows Recall to segment pricing more effectively across user needs.

Hi Recall team, I’ve been testing the Recall MCP integration with YouTube cards and found a useful enhancement opportunity. Recall already extracts the full YouTube transcript and lets me export it from the Recall interface. Through MCP, however, an AI client can currently access only selected transcript chunks through search or get_document_content. The MCP does expose timestamped reader chunks, alongside the generated notebook summary chunks, which is great. But get_document_content returns a limited, non-contiguous selection rather than the complete transcript. I can retrieve additional sections through targeted searches, but I cannot deterministically reconstruct or export the entire transcript. Would you consider adding a dedicated transcript operation, perhaps something like: get_transcript(card_id, format, include_timestamps, cursor/page_size) Ideally, it would: * Return the complete source transcript, separate from Recall’s summaries * Preserve timestamps * Support pagination for long videos * Maintain deterministic transcript order * Indicate whether the transcript is complete or partially available * Optionally return plain text, Markdown, WebVTT, or SRT This would make Recall particularly valuable as a source repository for AI research workflows. An assistant could retrieve the original transcript for detailed analysis, quotations, fact verification, or creating structured notes, while still using Recall’s summaries for faster discovery. I tested this with the YouTube card today. The MCP exposed transcript excerpts from approximately 00:00 through 14:36, but only as selected chunks rather than one complete, sequential transcript. Happy to share more details from the test if useful.

Integrate Amazon Kindle notes like Readwise

The "Add to notebook" sometimes comes with an option to save a chat to a new card. Often I want a chat response to be saved to a new card rather than added to the notebook on the current card. That'd be nice as a single click.

Could be a way to read all the document listening the reader.

trigger card created custom prompt to tag it with right tag from selected tags or create new one and create summary based on the prompt

...::: User story :::... As a regular Recall user with a reasonably large knowledge base, I want to be shown an occasional surprising connection between two or more items in my own content, so that I rediscover things I've forgotten and notice links I would never have searched for. ...::: Problem :::... Recall is excellent at cross-referencing when I go looking. The retrieval model is pull, not push. Everything I get out depends on me remembering that something is in there, or thinking to ask about it. With a few thousand cards accumulated this year, most of the base is effectively dormant. The value sitting in older material is invisible unless a search query happens to hit it. Search rewards what I already know I want. It does nothing for what I've forgotten I have. ...::: Proposed behaviour :::... On login, or on page load with sensible frequency capping, surface a small non-blocking card. It presents one insight drawn from my own knowledge base. Framings could rotate: ■ Did you know - a fact, claim or observation surfaced from across the knowledge base and its connections. Not restricted by age or last-opened date. The test is whether it's interesting, not whether it's old. ■ Hidden gem - content saved but never revisited. ■ Unexpected connection - two cards from different topics, sources or time periods that share a concept, argument or contradiction. ■ Then and now - something saved a year ago that relates directly to something saved this week. ■ Contradiction watch - two sources in the base that disagree on the same point. One says a control works, another says it doesn't. I may have saved both without noticing they conflict, and holding both unexamined is a small blind spot worth exposing. ■ Bridge card - a single item that sits between two otherwise unconnected clusters. It's the only thing linking two areas of interest, which usually means either it's unusually valuable or the connection is worth developing. ■ Orphan cluster - a group of related cards that connects to nothing else in the base. A topic that was picked up, explored briefly, and abandoned. Useful for spotting dropped threads. ■ Recurring theme - a concept that keeps appearing across sources I never filed together, saved over months without ever being named as a topic. Effectively: here's a pattern in your reading that you haven't noticed you have. ■ Source echo - two different authors, publications or formats making substantially the same argument. Either corroboration worth trusting more, or a sign that a single idea is circulating unchallenged and both sources trace back to one origin. ...::: Requirements :::... ■ Insights are drawn only from my own content. No external suggestions. ■ Each insight links directly to the source cards so it can be followed up immediately. ■ Frequency is configurable: every load, daily, weekly, off. ■ Fully dismissible and fully disableable, in settings, not buried. ■ Dismissal is per-insight, not per-session, so a bad one doesn't cost the feature its credibility. ■ Feedback signal on each card, something as light as useful or not useful, to tune what gets surfaced. ■ Insight types should be individually toggleable. Some users will want connections but not trivia. ■ Where an insight type depends on age or visit history, prefer older and less-visited material. This does not apply to “Did you know”, which should draw from the whole base. ...::: Request motivation :::.... The reason I build my personal knowledge base is the belief that past reading compounds. In practice it only compounds if it resurfaces. This feature turns the archive from a lookup system into something that occasionally teaches me something about my own thinking. It's also a differentiator. Plenty of tools store and retrieve. Very few tell you something you didn't know you knew.